AGarioud commited on
Commit
16ea832
1 Parent(s): f7c4ec6

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +96 -26
README.md CHANGED
@@ -18,13 +18,15 @@ tags:
18
 
19
  ## Context & Data
20
  <hr style='margin-top:-1em; margin-bottom:0' />
21
- The hereby FLAIR (#2) dataset is sampled countrywide and is composed of over 20 billion annotated pixels of very high resolution aerial imagery at 0.2 m spatial resolution, acquired over three years and different months (spatio-temporal domains).
22
  Aerial imagery patches consist of 5 channels (RVB-Near Infrared-Elevation) and have corresponding annotation (with 19 semantic classes or 13 for the baselines).
23
  Furthermore, to integrate broader spatial context and temporal information, high resolution Sentinel-2 satellite 1-year time series with 10 spectral band are also provided.
24
  More than 50,000 Sentinel-2 acquisitions with 10 m spatial resolution are available.
25
  <br>
26
 
27
- The dataset covers 50 spatial domains, encompassing 916 areas spanning 817 km². This dataset provides a robust foundation for advancing land cover mapping techniques.<br><br>
 
 
28
 
29
  <style type="text/css">
30
  .tg {border-collapse:collapse;border-spacing:0;}
@@ -61,11 +63,13 @@ The dataset covers 50 spatial domains, encompassing 916 areas spanning 817 km².
61
  <th class="tg-zv4m"></th>
62
  <th class="tg-zv4m">Class</th>
63
  <th class="tg-8jgo">Freq.-train (%)</th>
64
- <th class="tg-8jgo">Freq.-test (%)</th>
 
65
  <th class="tg-zv4m"></th>
66
  <th class="tg-zv4m">Class</th>
67
  <th class="tg-8jgo">Freq.-train (%)</th>
68
- <th class="tg-8jgo">Freq.-test (%)</th>
 
69
  </tr>
70
  </thead>
71
  <tbody>
@@ -73,96 +77,115 @@ The dataset covers 50 spatial domains, encompassing 916 areas spanning 817 km².
73
  <td class="tg-2e1p"></td>
74
  <td class="tg-km2t">(1) Building</td>
75
  <td class="tg-8jgo">8.14</td>
 
76
  <td class="tg-8jgo">3.26</td>
77
  <td class="tg-l5fa"></td>
78
  <td class="tg-km2t">(11) Agricultural Land</td>
79
  <td class="tg-8jgo">10.98</td>
 
80
  <td class="tg-8jgo">18.19</td>
81
  </tr>
82
  <tr>
83
  <td class="tg-9efv"></td>
84
  <td class="tg-km2t">(2) Pervious surface</td>
85
  <td class="tg-8jgo">8.25</td>
 
86
  <td class="tg-8jgo">3.82</td>
87
  <td class="tg-rime"></td>
88
  <td class="tg-km2t">(12) Plowed land</td>
89
  <td class="tg-8jgo">3.88</td>
 
90
  <td class="tg-8jgo">1.81</td>
91
  </tr>
92
  <tr>
93
  <td class="tg-3m6m"></td>
94
  <td class="tg-km2t">(3) Impervious surface</td>
95
  <td class="tg-8jgo">13.72</td>
 
96
  <td class="tg-8jgo">5.87</td>
97
  <td class="tg-2cns"></td>
98
  <td class="tg-km2t">(13) Swimming pool</td>
99
  <td class="tg-8jgo">0.01</td>
 
100
  <td class="tg-8jgo">0.02</td>
101
  </tr>
102
  <tr>
103
  <td class="tg-r3rw"></td>
104
  <td class="tg-km2t">(4) Bare soil</td>
105
  <td class="tg-8jgo">3.47</td>
 
106
  <td class="tg-8jgo">1.6</td>
107
  <td class="tg-jjsp"></td>
108
  <td class="tg-km2t">(14) Snow</td>
109
  <td class="tg-8jgo">0.15</td>
 
110
  <td class="tg-8jgo">-</td>
111
  </tr>
112
  <tr>
113
  <td class="tg-9xgv"></td>
114
  <td class="tg-km2t">(5) Water</td>
115
  <td class="tg-8jgo">4.88</td>
 
116
  <td class="tg-8jgo">3.17</td>
117
  <td class="tg-2w6m"></td>
118
  <td class="tg-km2t">(15) Clear cut</td>
119
  <td class="tg-8jgo">0.15</td>
 
120
  <td class="tg-8jgo">0.82</td>
121
  </tr>
122
  <tr>
123
  <td class="tg-b45e"></td>
124
  <td class="tg-km2t">(6) Coniferous</td>
125
  <td class="tg-8jgo">2.74</td>
 
126
  <td class="tg-8jgo">10.24</td>
127
  <td class="tg-nla7"></td>
128
  <td class="tg-km2t">(16) Mixed</td>
129
  <td class="tg-8jgo">0.05</td>
 
130
  <td class="tg-8jgo">0.12</td>
131
  </tr>
132
  <tr>
133
  <td class="tg-qg2z"></td>
134
  <td class="tg-km2t">(7) Deciduous</td>
135
  <td class="tg-8jgo">15.38</td>
 
136
  <td class="tg-8jgo">24.79</td>
137
  <td class="tg-nv8o"></td>
138
  <td class="tg-km2t">(17) Ligneous</td>
139
  <td class="tg-8jgo">0.01</td>
 
140
  <td class="tg-8jgo">-</td>
141
  </tr>
142
  <tr>
143
  <td class="tg-grz5"></td>
144
  <td class="tg-km2t">(8) Brushwood</td>
145
  <td class="tg-8jgo">6.95</td>
146
- <td class="tg-8jgo">3.81</td>
 
147
  <td class="tg-bja1"></td>
148
  <td class="tg-km2t">(18) Greenhouse</td>
149
  <td class="tg-8jgo">0.12</td>
 
150
  <td class="tg-8jgo">0.15</td>
151
  </tr>
152
  <tr>
153
  <td class="tg-69kt"></td>
154
  <td class="tg-km2t">(9) Vineyard</td>
155
  <td class="tg-8jgo">3.13</td>
 
156
  <td class="tg-8jgo">2.55</td>
157
  <td class="tg-nto1"></td>
158
  <td class="tg-km2t">(19) Other</td>
159
  <td class="tg-8jgo">0.14</td>
 
160
  <td class="tg-8jgo">0.04</td>
161
  </tr>
162
  <tr>
163
  <td class="tg-r1r4"></td>
164
  <td class="tg-km2t">(10) Herbaceous vegetation</td>
165
  <td class="tg-8jgo">17.84</td>
 
166
  <td class="tg-8jgo">19.76</td>
167
  <td class="tg-zv4m"></td>
168
  <td class="tg-zv4m"></td>
@@ -177,8 +200,10 @@ The dataset covers 50 spatial domains, encompassing 916 areas spanning 817 km².
177
 
178
  ## Dataset Structure
179
  <hr style='margin-top:-1em; margin-bottom:0' />
180
- The FLAIR dataset consists of 77 762 patches. Each patch includes a high-resolution aerial image (512x512) at 0.2 m, a yearly satellite image time series (40x40 by default by wider areas are provided) with a spatial resolution of 10 m
181
- and associated cloud and snow masks, and pixel-precise elevation and land cover annotations at 0.2 m resolution (512x512).
 
 
182
 
183
  <p align="center"><img src="readme_imgs/flair-patches.png" alt="" style="width:70%;max-width:600px;"/></p><br>
184
 
@@ -219,21 +244,23 @@ Satellite
219
  Each pixel has been manually annotated by photo-interpretation of the 20 cm resolution aerial imagery, carried out by a team supervised by geography experts from the IGN.
220
  Movable objects like cars or boats are annotated according to their underlying cover.
221
 
222
- ### Training Splits
223
- The dataset is made up of 50 distinct spatial domains, aligned with the administrative boundaries of the French départements.
224
- For our experiments, we designate 32 domains for training, 8 for validation, and reserve 10 as the official test set.
 
225
  This arrangement ensures a balanced distribution of semantic classes, radiometric attributes, bioclimatic conditions, and acquisition times across each set.
226
  Consequently, every split accurately reflects the landscape diversity inherent to metropolitan France.
227
- It is important to mention that the patches come with meta-data permitting alternative splitting schemes, for example focused on domain shifts.
 
228
 
229
  Official domain split: <br/>
230
 
231
  <div style="display: flex; flex-wrap: nowrap; align-items: center">
232
  <div style="flex: 40%;">
233
- <img src="readme_imgs/flair-splits.png" alt="Your Image">
234
  </div>
235
 
236
- <div style="flex: 60%;">
237
  <table border="1">
238
  <tr>
239
  <th><font color="#c7254e">TRAIN:</font></th>
@@ -244,24 +271,53 @@ Official domain split: <br/>
244
  <td>D004, D014, D029, D031, D058, D066, D067, D077</td>
245
  </tr>
246
  <tr>
247
- <th><font color="#c7254e">TEST:</font></th>
248
- <td>D015, D022, D026, D036, D061, D064, D068, D069, D071, D084</td>
249
  </tr>
 
 
 
 
250
  </table>
251
-
252
  </div>
253
  </div>
254
 
255
  <br><br>
256
 
 
257
  ## Baseline code
258
  <hr style='margin-top:-1em; margin-bottom:0' />
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
259
  We propose the U-T&T model, a two-branch architecture that combines spatial and temporal information from very high-resolution aerial images and high-resolution satellite images into a single output. The U-Net architecture is employed for the spatial/texture branch, using a ResNet34 backbone model pre-trained on ImageNet. For the spatio-temporal branch,
260
  the U-TAE architecture incorporates a Temporal self-Attention Encoder (TAE) to explore the spatial and temporal characteristics of the Sentinel-2 time series data,
261
  applying attention masks at different resolutions during decoding. This model allows for the fusion of learned information from both sources,
262
  enhancing the representation of mono-date and time series data.
263
 
264
- U-T&T code repository &#128193; : https://github.com/IGNF/FLAIR-2-AI-Challenge <br/>
 
265
 
266
  <th><font color="#c7254e"><b>IMPORTANT!</b></font></th> <b>The structure of the current dataset differs from the one that comes with the GitHub repository.</b>
267
  To work with the current dataset, you need to replace the <font color=‘#D7881C’><em>src/load_data.py</em></font> file with the one provided here.
@@ -273,14 +329,8 @@ domains_train : ["D006_2020","D007_2020","D008_2019","D009_2019","D013_2020","D0
273
  domains_val : ["D004_2021","D014_2020","D029_2021","D031_2019","D058_2020","D066_2021","D067_2021","D077_2021"]
274
  domains_test : ["D015_2020","D022_2021","D026_2020","D036_2020","D061_2020","D064_2021","D068_2021","D069_2020","D071_2020","D084_2021"]
275
  ```
276
-
277
-
278
- <br><br>
279
-
280
-
281
- ## Reference
282
- <hr style='margin-top:-1em; margin-bottom:0' />
283
- Please include a citation to the following article if you use the FLAIR dataset:
284
 
285
  ```
286
  @inproceedings{garioud2023flair,
@@ -291,8 +341,28 @@ Please include a citation to the following article if you use the FLAIR dataset:
291
  doi={https://doi.org/10.48550/arXiv.2310.13336},
292
  }
293
  ```
 
294
 
295
- Link to the paper : https://arxiv.org/abs/2310.13336 <br>
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
296
 
297
  ## Acknowledgment
298
  <hr style='margin-top:-1em; margin-bottom:0' />
 
18
 
19
  ## Context & Data
20
  <hr style='margin-top:-1em; margin-bottom:0' />
21
+ The hereby FLAIR (#1 and #2) dataset is sampled countrywide and is composed of over 20 billion annotated pixels of very high resolution aerial imagery at 0.2 m spatial resolution, acquired over three years and different months (spatio-temporal domains).
22
  Aerial imagery patches consist of 5 channels (RVB-Near Infrared-Elevation) and have corresponding annotation (with 19 semantic classes or 13 for the baselines).
23
  Furthermore, to integrate broader spatial context and temporal information, high resolution Sentinel-2 satellite 1-year time series with 10 spectral band are also provided.
24
  More than 50,000 Sentinel-2 acquisitions with 10 m spatial resolution are available.
25
  <br>
26
 
27
+ The dataset covers 55 distinct spatial domains, encompassing 974 areas spanning 980 km². This dataset provides a robust foundation for advancing land cover mapping techniques.
28
+ We sample two test sets based on different input data and focus on semantic classes. The first test set (flair#1-test) uses very high resolution aerial imagery only and samples primarily anthropized land cover classes.
29
+ In contrast, the second test set (flair#2-test) combines aerial and satellite imagery and has more natural classes with temporal variations represented.<br><br>
30
 
31
  <style type="text/css">
32
  .tg {border-collapse:collapse;border-spacing:0;}
 
63
  <th class="tg-zv4m"></th>
64
  <th class="tg-zv4m">Class</th>
65
  <th class="tg-8jgo">Freq.-train (%)</th>
66
+ <th class="tg-8jgo">Freq.-test flair#1 (%)</th>
67
+ <th class="tg-8jgo">Freq.-test flair#2 (%)</th>
68
  <th class="tg-zv4m"></th>
69
  <th class="tg-zv4m">Class</th>
70
  <th class="tg-8jgo">Freq.-train (%)</th>
71
+ <th class="tg-8jgo">Freq.-test flair#1 (%)</th>
72
+ <th class="tg-8jgo">Freq.-test flair#2 (%)</th>
73
  </tr>
74
  </thead>
75
  <tbody>
 
77
  <td class="tg-2e1p"></td>
78
  <td class="tg-km2t">(1) Building</td>
79
  <td class="tg-8jgo">8.14</td>
80
+ <td class="tg-8jgo">8.6</td>
81
  <td class="tg-8jgo">3.26</td>
82
  <td class="tg-l5fa"></td>
83
  <td class="tg-km2t">(11) Agricultural Land</td>
84
  <td class="tg-8jgo">10.98</td>
85
+ <td class="tg-8jgo">6.95</td>
86
  <td class="tg-8jgo">18.19</td>
87
  </tr>
88
  <tr>
89
  <td class="tg-9efv"></td>
90
  <td class="tg-km2t">(2) Pervious surface</td>
91
  <td class="tg-8jgo">8.25</td>
92
+ <td class="tg-8jgo">7.34</td>
93
  <td class="tg-8jgo">3.82</td>
94
  <td class="tg-rime"></td>
95
  <td class="tg-km2t">(12) Plowed land</td>
96
  <td class="tg-8jgo">3.88</td>
97
+ <td class="tg-8jgo">2.25</td>
98
  <td class="tg-8jgo">1.81</td>
99
  </tr>
100
  <tr>
101
  <td class="tg-3m6m"></td>
102
  <td class="tg-km2t">(3) Impervious surface</td>
103
  <td class="tg-8jgo">13.72</td>
104
+ <td class="tg-8jgo">14.98</td>
105
  <td class="tg-8jgo">5.87</td>
106
  <td class="tg-2cns"></td>
107
  <td class="tg-km2t">(13) Swimming pool</td>
108
  <td class="tg-8jgo">0.01</td>
109
+ <td class="tg-8jgo">0.04</td>
110
  <td class="tg-8jgo">0.02</td>
111
  </tr>
112
  <tr>
113
  <td class="tg-r3rw"></td>
114
  <td class="tg-km2t">(4) Bare soil</td>
115
  <td class="tg-8jgo">3.47</td>
116
+ <td class="tg-8jgo">4.36</td>
117
  <td class="tg-8jgo">1.6</td>
118
  <td class="tg-jjsp"></td>
119
  <td class="tg-km2t">(14) Snow</td>
120
  <td class="tg-8jgo">0.15</td>
121
+ <td class="tg-8jgo">-</td>
122
  <td class="tg-8jgo">-</td>
123
  </tr>
124
  <tr>
125
  <td class="tg-9xgv"></td>
126
  <td class="tg-km2t">(5) Water</td>
127
  <td class="tg-8jgo">4.88</td>
128
+ <td class="tg-8jgo">5.98</td>
129
  <td class="tg-8jgo">3.17</td>
130
  <td class="tg-2w6m"></td>
131
  <td class="tg-km2t">(15) Clear cut</td>
132
  <td class="tg-8jgo">0.15</td>
133
+ <td class="tg-8jgo">0.01</td>
134
  <td class="tg-8jgo">0.82</td>
135
  </tr>
136
  <tr>
137
  <td class="tg-b45e"></td>
138
  <td class="tg-km2t">(6) Coniferous</td>
139
  <td class="tg-8jgo">2.74</td>
140
+ <td class="tg-8jgo">2.39</td>
141
  <td class="tg-8jgo">10.24</td>
142
  <td class="tg-nla7"></td>
143
  <td class="tg-km2t">(16) Mixed</td>
144
  <td class="tg-8jgo">0.05</td>
145
+ <td class="tg-8jgo">-</td>
146
  <td class="tg-8jgo">0.12</td>
147
  </tr>
148
  <tr>
149
  <td class="tg-qg2z"></td>
150
  <td class="tg-km2t">(7) Deciduous</td>
151
  <td class="tg-8jgo">15.38</td>
152
+ <td class="tg-8jgo">13.91</td>
153
  <td class="tg-8jgo">24.79</td>
154
  <td class="tg-nv8o"></td>
155
  <td class="tg-km2t">(17) Ligneous</td>
156
  <td class="tg-8jgo">0.01</td>
157
+ <td class="tg-8jgo">0.03</td>
158
  <td class="tg-8jgo">-</td>
159
  </tr>
160
  <tr>
161
  <td class="tg-grz5"></td>
162
  <td class="tg-km2t">(8) Brushwood</td>
163
  <td class="tg-8jgo">6.95</td>
164
+ <td class="tg-8jgo">6.91</td>
165
+ <td class="tg-8jgo">3.81</td>
166
  <td class="tg-bja1"></td>
167
  <td class="tg-km2t">(18) Greenhouse</td>
168
  <td class="tg-8jgo">0.12</td>
169
+ <td class="tg-8jgo">0.2</td>
170
  <td class="tg-8jgo">0.15</td>
171
  </tr>
172
  <tr>
173
  <td class="tg-69kt"></td>
174
  <td class="tg-km2t">(9) Vineyard</td>
175
  <td class="tg-8jgo">3.13</td>
176
+ <td class="tg-8jgo">3.87</td>
177
  <td class="tg-8jgo">2.55</td>
178
  <td class="tg-nto1"></td>
179
  <td class="tg-km2t">(19) Other</td>
180
  <td class="tg-8jgo">0.14</td>
181
+ <td class="tg-8jgo">0.-</td>
182
  <td class="tg-8jgo">0.04</td>
183
  </tr>
184
  <tr>
185
  <td class="tg-r1r4"></td>
186
  <td class="tg-km2t">(10) Herbaceous vegetation</td>
187
  <td class="tg-8jgo">17.84</td>
188
+ <td class="tg-8jgo">22.17</td>
189
  <td class="tg-8jgo">19.76</td>
190
  <td class="tg-zv4m"></td>
191
  <td class="tg-zv4m"></td>
 
200
 
201
  ## Dataset Structure
202
  <hr style='margin-top:-1em; margin-bottom:0' />
203
+ The FLAIR dataset consists of a total of 93 462 patches: 61 712 patches for the train/val dataset, 15 700 patches for flair#1-test and 16 050 patches for flair#2-test.
204
+
205
+ Each patch includes a high-resolution aerial image (512x512) at 0.2 m, a yearly satellite image time series (40x40 by default by wider areas are provided) with a spatial resolution of 10 m
206
+ and associated cloud and snow masks (available in train/val and flair#2-test), and pixel-precise elevation and land cover annotations at 0.2 m resolution (512x512).
207
 
208
  <p align="center"><img src="readme_imgs/flair-patches.png" alt="" style="width:70%;max-width:600px;"/></p><br>
209
 
 
244
  Each pixel has been manually annotated by photo-interpretation of the 20 cm resolution aerial imagery, carried out by a team supervised by geography experts from the IGN.
245
  Movable objects like cars or boats are annotated according to their underlying cover.
246
 
247
+ ### Data Splits
248
+ The dataset is made up of 55 distinct spatial domains, aligned with the administrative boundaries of the French départements.
249
+ For our experiments, we designate 32 domains for training, 8 for validation, and reserve 10 official test sets for flair#1-test and flair#2-test.
250
+ It can also be noted that some domains are common in the flair#1-test and flair#2-test datasets but cover different areas within the domain.
251
  This arrangement ensures a balanced distribution of semantic classes, radiometric attributes, bioclimatic conditions, and acquisition times across each set.
252
  Consequently, every split accurately reflects the landscape diversity inherent to metropolitan France.
253
+ It is important to mention that the patches come with meta-data permitting alternative splitting schemes.
254
+
255
 
256
  Official domain split: <br/>
257
 
258
  <div style="display: flex; flex-wrap: nowrap; align-items: center">
259
  <div style="flex: 40%;">
260
+ <img src="readme_imgs/flair-splits.png" alt="flair-splits">
261
  </div>
262
 
263
+ <div style="flex: 60%; margin: auto;"">
264
  <table border="1">
265
  <tr>
266
  <th><font color="#c7254e">TRAIN:</font></th>
 
271
  <td>D004, D014, D029, D031, D058, D066, D067, D077</td>
272
  </tr>
273
  <tr>
274
+ <th><font color="#c7254e">TEST-flair#1:</font></th>
275
+ <td>D012, D022, D026, D064, D068, D071, D075, D076, D083, D085</td>
276
  </tr>
277
+ <tr>
278
+ <th><font color="#c7254e">TEST-flair#2:</font></th>
279
+ <td>D015, D022, D026, D036, D061, D064, D068, D069, D071, D084</td>
280
+ </tr>
281
  </table>
 
282
  </div>
283
  </div>
284
 
285
  <br><br>
286
 
287
+
288
  ## Baseline code
289
  <hr style='margin-top:-1em; margin-bottom:0' />
290
+ <br>
291
+
292
+ ### Flair #1 (aerial only)
293
+ A U-Net architecture with a pre-trained ResNet34 encoder from the pytorch segmentation models library is used for the baselines.
294
+ The used architecture allows integration of patch-wise metadata information and employs commonly used image data augmentation techniques.
295
+
296
+ Flair#1 code repository &#128193; : https://github.com/IGNF/FLAIR-1<br/>
297
+ Link to the paper : https://arxiv.org/pdf/2211.12979.pdf <br>
298
+
299
+ Please include a citation to the following article if you use the FLAIR#1 dataset:
300
+
301
+ ```
302
+ @article{ign2022flair1,
303
+ doi = {10.13140/RG.2.2.30183.73128/1},
304
+ url = {https://arxiv.org/pdf/2211.12979.pdf},
305
+ author = {Garioud, Anatol and Peillet, Stéphane and Bookjans, Eva and Giordano, Sébastien and Wattrelos, Boris},
306
+ title = {FLAIR #1: semantic segmentation and domain adaptation dataset},
307
+ publisher = {arXiv},
308
+ year = {2022}
309
+ }
310
+ ```
311
+ <br>
312
+
313
+ ### Flair #2 (aerial and satellite)
314
  We propose the U-T&T model, a two-branch architecture that combines spatial and temporal information from very high-resolution aerial images and high-resolution satellite images into a single output. The U-Net architecture is employed for the spatial/texture branch, using a ResNet34 backbone model pre-trained on ImageNet. For the spatio-temporal branch,
315
  the U-TAE architecture incorporates a Temporal self-Attention Encoder (TAE) to explore the spatial and temporal characteristics of the Sentinel-2 time series data,
316
  applying attention masks at different resolutions during decoding. This model allows for the fusion of learned information from both sources,
317
  enhancing the representation of mono-date and time series data.
318
 
319
+ U-T&T code repository &#128193; : https://github.com/IGNF/FLAIR-2<br/>
320
+ Link to the paper : https://arxiv.org/abs/2310.13336 <br>
321
 
322
  <th><font color="#c7254e"><b>IMPORTANT!</b></font></th> <b>The structure of the current dataset differs from the one that comes with the GitHub repository.</b>
323
  To work with the current dataset, you need to replace the <font color=‘#D7881C’><em>src/load_data.py</em></font> file with the one provided here.
 
329
  domains_val : ["D004_2021","D014_2020","D029_2021","D031_2019","D058_2020","D066_2021","D067_2021","D077_2021"]
330
  domains_test : ["D015_2020","D022_2021","D026_2020","D036_2020","D061_2020","D064_2021","D068_2021","D069_2020","D071_2020","D084_2021"]
331
  ```
332
+ <br>
333
+ Please include a citation to the following article if you use the FLAIR#2 dataset:
 
 
 
 
 
 
334
 
335
  ```
336
  @inproceedings{garioud2023flair,
 
341
  doi={https://doi.org/10.48550/arXiv.2310.13336},
342
  }
343
  ```
344
+ <br>
345
 
346
+ ## CodaLab challenges
347
+ <hr style='margin-top:-1em; margin-bottom:0' />
348
+
349
+ The FLAIR dataset was used for two challenges organized by IGN in 2023 on the CodaLab platform.<br>
350
+ Challenge FLAIR#1 : https://codalab.lisn.upsaclay.fr/competitions/8769 <br>
351
+ Challenge FLAIR#2 : https://codalab.lisn.upsaclay.fr/competitions/13447 <br>
352
+ <br>
353
+
354
+ flair#1-test | The podium:
355
+ 🥇 businiao - 0.65920
356
+ 🥈 Breizhchess - 0.65600
357
+ 🥉 wangzhiyu918 - 0.64930
358
+
359
+ flair#2-test | The podium:
360
+ The podium:
361
+ 🥇 strakajk - 0.64130
362
+ 🥈 Breizhchess - 0.63550
363
+ 🥉 qwerty64 - 0.63510
364
+
365
+ <br>
366
 
367
  ## Acknowledgment
368
  <hr style='margin-top:-1em; margin-bottom:0' />